Face Generation¶

In this project, you'll use generative adversarial networks to generate new images of faces.

Get the Data¶

You'll be using two datasets in this project:

MNIST
CelebA

Since the celebA dataset is complex and you're doing GANs in a project for the first time, we want you to test your neural network on MNIST before CelebA. Running the GANs on MNIST will allow you to see how well your model trains sooner.

If you're using FloydHub, set data_dir to "/input" and use the FloydHub data ID "R5KrjnANiKVhLWAkpXhNBe".

#data_dir = './data'

# FloydHub - Use with data ID "R5KrjnANiKVhLWAkpXhNBe"
data_dir = '/input'


"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import helper

helper.download_extract('mnist', data_dir)
helper.download_extract('celeba', data_dir)

Found mnist Data
Found celeba Data

Explore the Data¶

MNIST¶

As you're aware, the MNIST dataset contains images of handwritten digits. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
%matplotlib inline
import os
from glob import glob
from matplotlib import pyplot

mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'mnist/*.jpg'))[:show_n_images], 28, 28, 'L')
pyplot.imshow(helper.images_square_grid(mnist_images, 'L'), cmap='gray')

<matplotlib.image.AxesImage at 0x7f21b2452400>

CelebA¶

The CelebFaces Attributes Dataset (CelebA) dataset contains over 200,000 celebrity images with annotations. Since you're going to be generating faces, you won't need the annotations. You can view the first number of examples by changing show_n_images.

show_n_images = 25

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
mnist_images = helper.get_batch(glob(os.path.join(data_dir, 'img_align_celeba/*.jpg'))[:show_n_images], 28, 28, 'RGB')
pyplot.imshow(helper.images_square_grid(mnist_images, 'RGB'))

<matplotlib.image.AxesImage at 0x7f21b2346ba8>

Preprocess the Data¶

Since the project's main focus is on building the GANs, we'll preprocess the data for you. The values of the MNIST and CelebA dataset will be in the range of -0.5 to 0.5 of 28x28 dimensional images. The CelebA images will be cropped to remove parts of the image that don't include a face, then resized down to 28x28.

The MNIST images are black and white images with a single color channel while the CelebA images have 3 color channels (RGB color channel).

Build the Neural Network¶

You'll build the components necessary to build a GANs by implementing the following functions below:

model_inputs
discriminator
generator
model_loss
model_opt
train

Check the Version of TensorFlow and Access to GPU¶

This will check to make sure you have the correct version of TensorFlow and access to a GPU

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
from distutils.version import LooseVersion
import warnings
import tensorflow as tf

# Check TensorFlow Version
assert LooseVersion(tf.__version__) >= LooseVersion('1.0'), 'Please use TensorFlow version 1.0 or newer.  You are using {}'.format(tf.__version__)
print('TensorFlow Version: {}'.format(tf.__version__))

# Check for a GPU
if not tf.test.gpu_device_name():
    warnings.warn('No GPU found. Please use a GPU to train your neural network.')
else:
    print('Default GPU Device: {}'.format(tf.test.gpu_device_name()))

TensorFlow Version: 1.1.0
Default GPU Device: /gpu:0

Input¶

Implement the model_inputs function to create TF Placeholders for the Neural Network. It should create the following placeholders:

Real input images placeholder with rank 4 using image_width, image_height, and image_channels.
Z input placeholder with rank 2 using z_dim.
Learning rate placeholder with rank 0.

Return the placeholders in the following the tuple (tensor of real input images, tensor of z data)

import problem_unittests as tests

def model_inputs(image_width, image_height, image_channels, z_dim):
    """
    Create the model inputs
    :param image_width: The input image width
    :param image_height: The input image height
    :param image_channels: The number of image channels
    :param z_dim: The dimension of Z
    :return: Tuple of (tensor of real input images, tensor of z data, learning rate)
    """
    inputs_real = tf.placeholder(tf.float32, (None, image_width, image_height, image_channels), name='input_real')
    inputs_z = tf.placeholder(tf.float32, (None, z_dim), name='input_z')
    learn_rate = tf.placeholder(tf.float32, name='learn_rate')

    return inputs_real, inputs_z, learn_rate


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_inputs(model_inputs)

Tests Passed

Discriminator¶

Implement discriminator to create a discriminator neural network that discriminates on images. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "discriminator" to allow the variables to be reused. The function should return a tuple of (tensor output of the discriminator, tensor logits of the discriminator).

def discriminator(images, reuse=False, alpha=0.2, keep_prob=1):
    """
    Create the discriminator network
    :param images: Tensor of input image(s)
    :param reuse: Boolean if the weights should be reused
    :return: Tuple of (tensor output of the discriminator, tensor logits of the discriminator)
    """
    with tf.variable_scope('discriminator', reuse=reuse):     
        # Input layer is 28x28x3/1
        x1 = tf.layers.conv2d(images, 64, 5, strides=2, padding='same')
        relu1 = tf.maximum(alpha * x1, x1)
        # 14x14x64
        
        x2 = tf.layers.conv2d(relu1, 128, 5, strides=2, padding='same')
        drop2 = tf.nn.dropout(x2, keep_prob)
        bn2 = tf.layers.batch_normalization(drop2, training=True)
        relu2 = tf.maximum(alpha * bn2, bn2)
        # 7x7x128
        
        x3 = tf.layers.conv2d(relu2, 256, 5, strides=1, padding='same')
        drop3 = tf.nn.dropout(x3, keep_prob)
        bn3 = tf.layers.batch_normalization(drop3, training=True)
        relu3 = tf.maximum(alpha * bn3, bn3)
        # 7x7x256

        # Flatten it
        flat = tf.reshape(relu3, (-1, 7*7*256))
        logits = tf.layers.dense(flat, 1)
        out = tf.sigmoid(logits)
        
        return out, logits

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_discriminator(discriminator, tf)

Tests Passed

Generator¶

Implement generator to generate an image using z. This function should be able to reuse the variables in the neural network. Use tf.variable_scope with a scope name of "generator" to allow the variables to be reused. The function should return the generated 28 x 28 x out_channel_dim images.

def generator(z, out_channel_dim, is_train=True, alpha=0.2):
    """
    Create the generator network
    :param z: Input z
    :param out_channel_dim: The number of channels in the output image
    :param is_train: Boolean if generator is being used for training
    :return: The tensor output of the generator
    """
    with tf.variable_scope('generator', reuse=not is_train):
        # First fully connected layer
        x1 = tf.layers.dense(z, 7*7*512)
        # Reshape it to start the convolutional stack
        x1 = tf.reshape(x1, (-1, 7, 7, 512))
        x1 = tf.layers.batch_normalization(x1, training=is_train)
        x1 = tf.maximum(alpha * x1, x1)
        # 7x7x512 now
        
        x2 = tf.layers.conv2d_transpose(x1, 256, 5, strides=2, padding='same')
        x2 = tf.layers.batch_normalization(x2, training=is_train)
        x2 = tf.maximum(alpha * x2, x2)
        # 14x14x256 now
        
        x3 = tf.layers.conv2d_transpose(x2, 128, 5, strides=2, padding='same')
        x3 = tf.layers.batch_normalization(x3, training=is_train)
        x3 = tf.maximum(alpha * x3, x3)
        # 28x28x128 now
        
        # Output layer
        logits = tf.layers.conv2d_transpose(x3, out_channel_dim, 5, strides=1, padding='SAME')
        # 28x28x3 now
        
        out = tf.tanh(logits)
        
        return out

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_generator(generator, tf)

Tests Passed

Loss¶

Implement model_loss to build the GANs for training and calculate the loss. The function should return a tuple of (discriminator loss, generator loss). Use the following functions you implemented:

discriminator(images, reuse=False)
generator(z, out_channel_dim, is_train=True)

def model_loss(input_real, input_z, out_channel_dim, alpha=0.2, keep_p=1):
    """
    Get the loss for the discriminator and generator
    :param input_real: Images from the real dataset
    :param input_z: Z input
    :param out_channel_dim: The number of channels in the output image
    :return: A tuple of (discriminator loss, generator loss)
    """
    g_model = generator(input_z, out_channel_dim, alpha=alpha)
    d_model_real, d_logits_real = discriminator(input_real, alpha=alpha, keep_prob=keep_p)
    d_model_fake, d_logits_fake = discriminator(g_model, reuse=True, alpha=alpha, keep_prob=keep_p)

    d_loss_real = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_real, labels=tf.ones_like(d_model_real)))
    d_loss_fake = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.zeros_like(d_model_fake)))
    g_loss = tf.reduce_mean(
        tf.nn.sigmoid_cross_entropy_with_logits(logits=d_logits_fake, labels=tf.ones_like(d_model_fake)))

    d_loss = d_loss_real + d_loss_fake
    
    return d_loss, g_loss


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_loss(model_loss)

Tests Passed

Optimization¶

Implement model_opt to create the optimization operations for the GANs. Use tf.trainable_variables to get all the trainable variables. Filter the variables with names that are in the discriminator and generator scope names. The function should return a tuple of (discriminator training operation, generator training operation).

def model_opt(d_loss, g_loss, learning_rate, beta1):
    """
    Get optimization operations
    :param d_loss: Discriminator loss Tensor
    :param g_loss: Generator loss Tensor
    :param learning_rate: Learning Rate Placeholder
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :return: A tuple of (discriminator training operation, generator training operation)
    """
    t_vars = tf.trainable_variables()
    d_vars = [var for var in t_vars if var.name.startswith('discriminator')]
    g_vars = [var for var in t_vars if var.name.startswith('generator')]

    # Optimize
    with tf.control_dependencies(tf.get_collection(tf.GraphKeys.UPDATE_OPS)):
        d_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(d_loss, var_list=d_vars)
        g_train_opt = tf.train.AdamOptimizer(learning_rate, beta1=beta1).minimize(g_loss, var_list=g_vars)

    return d_train_opt, g_train_opt


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
tests.test_model_opt(model_opt, tf)

Tests Passed

Neural Network Training¶

Show Output¶

Use this function to show the current output of the generator during training. It will help you determine how well the GANs is training.

"""
DON'T MODIFY ANYTHING IN THIS CELL
"""
import numpy as np

def show_generator_output(sess, n_images, input_z, out_channel_dim, image_mode):
    """
    Show example output for the generator
    :param sess: TensorFlow session
    :param n_images: Number of Images to display
    :param input_z: Input Z Tensor
    :param out_channel_dim: The number of channels in the output image
    :param image_mode: The mode to use for images ("RGB" or "L")
    """
    cmap = None if image_mode == 'RGB' else 'gray'
    z_dim = input_z.get_shape().as_list()[-1]
    example_z = np.random.uniform(-1, 1, size=[n_images, z_dim])

    samples = sess.run(
        generator(input_z, out_channel_dim, False),
        feed_dict={input_z: example_z})

    images_grid = helper.images_square_grid(samples, image_mode)
    pyplot.imshow(images_grid, cmap=cmap)
    pyplot.show()

Train¶

Implement train to build and train the GANs. Use the following functions you implemented:

model_inputs(image_width, image_height, image_channels, z_dim)
model_loss(input_real, input_z, out_channel_dim)
model_opt(d_loss, g_loss, learning_rate, beta1)

Use the show_generator_output to show generator output while you train. Running show_generator_output for every batch will drastically increase training time and increase the size of the notebook. It's recommended to print the generator output every 100 batches.

def scale(x, feature_range=(-1, 1)):
    # scale to (0, 1)
    x = ((x - x.min())/(0.5 - x.min()))
    
    # scale to feature_range
    min, max = feature_range
    x = x * (max - min) + min
    return x

def train(epoch_count, batch_size, z_dim, learning_rate, beta1, get_batches, data_shape, data_image_mode, keep_prob):
    """
    Train the GAN
    :param epoch_count: Number of epochs
    :param batch_size: Batch Size
    :param z_dim: Z dimension
    :param learning_rate: Learning Rate
    :param beta1: The exponential decay rate for the 1st moment in the optimizer
    :param get_batches: Function to get batches
    :param data_shape: Shape of the data
    :param data_image_mode: The image mode to use for images ("RGB" or "L")
    """
    #tf.reset_default_graph()
    image_dim = data_shape[1:]
    out_channel_dim = data_shape[3]
    
    keep_p = tf.placeholder(tf.float32, name='keep_b')
    input_real, input_z, learn_rate = model_inputs(*image_dim, z_dim=z_dim)
        
    d_loss, g_loss = model_loss(input_real, input_z, out_channel_dim, alpha=0.2, keep_p=keep_p)
        
    d_opt, g_opt = model_opt(d_loss, g_loss, learning_rate, beta1)
    
    step = 0
    with tf.Session() as sess:
        sess.run(tf.global_variables_initializer())
        for epoch_i in range(epoch_count):
            
            # Train 2 epochs per time
            if epoch_i % 2 == 0:
                resp = input('Train more 2 epochs? [y/n]')
                if resp == 'y':
                    pass
                if resp == 'n':
                    show_generator_output(sess, 16, input_z, out_channel_dim, data_image_mode)
                    return
    
            for batch_images in get_batches(batch_size):
                step += 1
                # Sample random noise for G
                batch_z = np.random.uniform(-1, 1, size=(batch_size, z_dim))
                
                # Rescale real image input to (-1, 1)
                batch_images = scale(batch_images)

                # Run optimizers
                _ = sess.run(d_opt, feed_dict={input_real: batch_images, input_z: batch_z, keep_p: keep_prob})
                _ = sess.run(g_opt, feed_dict={input_z: batch_z, input_real: batch_images, keep_p: keep_prob})
                
                if step % 10 == 0:
                    train_loss_d = d_loss.eval({input_z: batch_z, input_real: batch_images, keep_p: 1.})
                    train_loss_g = g_loss.eval({input_z: batch_z, keep_p: 1.})
                    
                    print("Epoch {}/{}...".format(epoch_i+1, epoch_count),
                          "Discriminator Loss: {:.4f}...".format(train_loss_d),
                          "Generator Loss: {:.4f}".format(train_loss_g))

                if step % 100 == 0:
                    show_generator_output(sess, 16, input_z, out_channel_dim, data_image_mode)
                    #plt.show()

MNIST¶

Test your GANs architecture on MNIST. After 2 epochs, the GANs should be able to generate images that look like handwritten digits. Make sure the loss of the generator is lower than the loss of the discriminator or close to 0.

batch_size = 64
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5
keep_prob = 1.0


"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 1

mnist_dataset = helper.Dataset('mnist', glob(os.path.join(data_dir, 'mnist/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, mnist_dataset.get_batches,
          mnist_dataset.shape, mnist_dataset.image_mode, keep_prob)

Epoch 1/1... Discriminator Loss: 0.1356... Generator Loss: 5.4989
Epoch 1/1... Discriminator Loss: 0.0448... Generator Loss: 4.5351
Epoch 1/1... Discriminator Loss: 5.9368... Generator Loss: 0.0054
Epoch 1/1... Discriminator Loss: 2.9559... Generator Loss: 0.3691
Epoch 1/1... Discriminator Loss: 1.3718... Generator Loss: 0.8958
Epoch 1/1... Discriminator Loss: 1.2327... Generator Loss: 0.8789
Epoch 1/1... Discriminator Loss: 1.0919... Generator Loss: 0.8178
Epoch 1/1... Discriminator Loss: 1.2184... Generator Loss: 0.9122
Epoch 1/1... Discriminator Loss: 1.2368... Generator Loss: 1.5033
Epoch 1/1... Discriminator Loss: 1.8424... Generator Loss: 0.2734

Epoch 1/1... Discriminator Loss: 1.3256... Generator Loss: 0.5985
Epoch 1/1... Discriminator Loss: 1.1964... Generator Loss: 0.7991
Epoch 1/1... Discriminator Loss: 1.1705... Generator Loss: 1.2143
Epoch 1/1... Discriminator Loss: 1.2393... Generator Loss: 0.5625
Epoch 1/1... Discriminator Loss: 1.3262... Generator Loss: 1.3087
Epoch 1/1... Discriminator Loss: 1.3215... Generator Loss: 1.3607
Epoch 1/1... Discriminator Loss: 1.0414... Generator Loss: 0.9684
Epoch 1/1... Discriminator Loss: 1.1028... Generator Loss: 0.6467
Epoch 1/1... Discriminator Loss: 1.1322... Generator Loss: 0.7206
Epoch 1/1... Discriminator Loss: 1.2341... Generator Loss: 0.4855

Epoch 1/1... Discriminator Loss: 0.9239... Generator Loss: 1.1272
Epoch 1/1... Discriminator Loss: 1.0113... Generator Loss: 1.6246
Epoch 1/1... Discriminator Loss: 0.8876... Generator Loss: 1.2344
Epoch 1/1... Discriminator Loss: 0.9755... Generator Loss: 1.3057
Epoch 1/1... Discriminator Loss: 0.9381... Generator Loss: 0.7546
Epoch 1/1... Discriminator Loss: 0.9597... Generator Loss: 1.2170
Epoch 1/1... Discriminator Loss: 1.4057... Generator Loss: 0.4268
Epoch 1/1... Discriminator Loss: 1.4496... Generator Loss: 0.4996
Epoch 1/1... Discriminator Loss: 1.1408... Generator Loss: 0.8016
Epoch 1/1... Discriminator Loss: 1.1450... Generator Loss: 0.6383

Epoch 1/1... Discriminator Loss: 1.1943... Generator Loss: 0.8572
Epoch 1/1... Discriminator Loss: 1.2407... Generator Loss: 0.9396
Epoch 1/1... Discriminator Loss: 1.5313... Generator Loss: 0.3064
Epoch 1/1... Discriminator Loss: 0.9909... Generator Loss: 1.0395
Epoch 1/1... Discriminator Loss: 1.4354... Generator Loss: 0.3650
Epoch 1/1... Discriminator Loss: 1.0404... Generator Loss: 1.1439
Epoch 1/1... Discriminator Loss: 1.3595... Generator Loss: 0.4633
Epoch 1/1... Discriminator Loss: 0.9102... Generator Loss: 0.9128
Epoch 1/1... Discriminator Loss: 1.0535... Generator Loss: 0.9435
Epoch 1/1... Discriminator Loss: 0.9827... Generator Loss: 1.3009

Epoch 1/1... Discriminator Loss: 1.0003... Generator Loss: 0.8813
Epoch 1/1... Discriminator Loss: 1.2062... Generator Loss: 0.5341
Epoch 1/1... Discriminator Loss: 1.2246... Generator Loss: 0.5118
Epoch 1/1... Discriminator Loss: 1.1317... Generator Loss: 1.2511
Epoch 1/1... Discriminator Loss: 1.0873... Generator Loss: 0.6953
Epoch 1/1... Discriminator Loss: 1.2422... Generator Loss: 0.5060
Epoch 1/1... Discriminator Loss: 0.8704... Generator Loss: 0.8376
Epoch 1/1... Discriminator Loss: 1.0928... Generator Loss: 0.8741
Epoch 1/1... Discriminator Loss: 1.2214... Generator Loss: 0.5741
Epoch 1/1... Discriminator Loss: 0.9887... Generator Loss: 0.8290

Epoch 1/1... Discriminator Loss: 1.0582... Generator Loss: 1.4704
Epoch 1/1... Discriminator Loss: 1.1718... Generator Loss: 0.6368
Epoch 1/1... Discriminator Loss: 0.9292... Generator Loss: 0.9499
Epoch 1/1... Discriminator Loss: 1.2321... Generator Loss: 0.5544
Epoch 1/1... Discriminator Loss: 1.3573... Generator Loss: 0.4299
Epoch 1/1... Discriminator Loss: 1.1349... Generator Loss: 0.6974
Epoch 1/1... Discriminator Loss: 1.3742... Generator Loss: 0.4166
Epoch 1/1... Discriminator Loss: 1.1038... Generator Loss: 0.7770
Epoch 1/1... Discriminator Loss: 0.9960... Generator Loss: 0.8260
Epoch 1/1... Discriminator Loss: 1.0765... Generator Loss: 0.8365

Epoch 1/1... Discriminator Loss: 1.0932... Generator Loss: 0.6556
Epoch 1/1... Discriminator Loss: 1.1733... Generator Loss: 1.2166
Epoch 1/1... Discriminator Loss: 1.1344... Generator Loss: 0.6603
Epoch 1/1... Discriminator Loss: 1.1106... Generator Loss: 1.3474
Epoch 1/1... Discriminator Loss: 1.0234... Generator Loss: 0.7752
Epoch 1/1... Discriminator Loss: 1.1092... Generator Loss: 0.6917
Epoch 1/1... Discriminator Loss: 1.0653... Generator Loss: 0.9978
Epoch 1/1... Discriminator Loss: 1.2277... Generator Loss: 0.4995
Epoch 1/1... Discriminator Loss: 0.9264... Generator Loss: 0.7869
Epoch 1/1... Discriminator Loss: 0.9360... Generator Loss: 1.1303

Epoch 1/1... Discriminator Loss: 0.9542... Generator Loss: 1.4096
Epoch 1/1... Discriminator Loss: 0.9456... Generator Loss: 0.8581
Epoch 1/1... Discriminator Loss: 1.0104... Generator Loss: 1.0380
Epoch 1/1... Discriminator Loss: 0.8691... Generator Loss: 1.2129
Epoch 1/1... Discriminator Loss: 1.9091... Generator Loss: 0.2101
Epoch 1/1... Discriminator Loss: 1.0213... Generator Loss: 0.9997
Epoch 1/1... Discriminator Loss: 1.5355... Generator Loss: 2.1744
Epoch 1/1... Discriminator Loss: 1.2729... Generator Loss: 0.4788
Epoch 1/1... Discriminator Loss: 1.4228... Generator Loss: 0.3824
Epoch 1/1... Discriminator Loss: 1.0814... Generator Loss: 1.5442

Epoch 1/1... Discriminator Loss: 1.0692... Generator Loss: 1.4833
Epoch 1/1... Discriminator Loss: 0.9403... Generator Loss: 1.3560
Epoch 1/1... Discriminator Loss: 1.0692... Generator Loss: 0.6854
Epoch 1/1... Discriminator Loss: 0.9444... Generator Loss: 0.9147
Epoch 1/1... Discriminator Loss: 0.9338... Generator Loss: 0.8397
Epoch 1/1... Discriminator Loss: 0.9797... Generator Loss: 0.7233
Epoch 1/1... Discriminator Loss: 1.2282... Generator Loss: 0.4945
Epoch 1/1... Discriminator Loss: 1.3732... Generator Loss: 0.5046
Epoch 1/1... Discriminator Loss: 1.0166... Generator Loss: 0.7315
Epoch 1/1... Discriminator Loss: 0.9324... Generator Loss: 1.0614

Epoch 1/1... Discriminator Loss: 0.9284... Generator Loss: 1.0279
Epoch 1/1... Discriminator Loss: 2.2660... Generator Loss: 2.1880
Epoch 1/1... Discriminator Loss: 1.2195... Generator Loss: 0.5255

CelebA¶

Run your GANs on CelebA. It will take around 20 minutes on the average GPU to run one epoch. You can run the whole epoch or stop when it starts to generate realistic faces.

batch_size = 64
z_dim = 100
learning_rate = 0.0002
beta1 = 0.5
keep_prob = 1.0

"""
DON'T MODIFY ANYTHING IN THIS CELL THAT IS BELOW THIS LINE
"""
epochs = 8

celeba_dataset = helper.Dataset('celeba', glob(os.path.join(data_dir, 'img_align_celeba/*.jpg')))
with tf.Graph().as_default():
    train(epochs, batch_size, z_dim, learning_rate, beta1, celeba_dataset.get_batches,
          celeba_dataset.shape, celeba_dataset.image_mode, keep_prob)

Train more 2 epochs? [y/n]y
Epoch 1/8... Discriminator Loss: 0.8272... Generator Loss: 1.3410
Epoch 1/8... Discriminator Loss: 0.6979... Generator Loss: 1.2693
Epoch 1/8... Discriminator Loss: 0.6438... Generator Loss: 7.7162
Epoch 1/8... Discriminator Loss: 0.2232... Generator Loss: 2.3780
Epoch 1/8... Discriminator Loss: 1.2383... Generator Loss: 0.6929
Epoch 1/8... Discriminator Loss: 0.8620... Generator Loss: 3.0325
Epoch 1/8... Discriminator Loss: 0.8726... Generator Loss: 2.5915
Epoch 1/8... Discriminator Loss: 0.8393... Generator Loss: 3.9013
Epoch 1/8... Discriminator Loss: 0.5531... Generator Loss: 2.1973
Epoch 1/8... Discriminator Loss: 0.4210... Generator Loss: 2.2769

Epoch 1/8... Discriminator Loss: 0.2407... Generator Loss: 4.0527
Epoch 1/8... Discriminator Loss: 1.2847... Generator Loss: 5.9022
Epoch 1/8... Discriminator Loss: 0.3634... Generator Loss: 4.7262
Epoch 1/8... Discriminator Loss: 0.4925... Generator Loss: 7.6501
Epoch 1/8... Discriminator Loss: 0.4744... Generator Loss: 1.5086
Epoch 1/8... Discriminator Loss: 0.8719... Generator Loss: 0.9727
Epoch 1/8... Discriminator Loss: 0.5643... Generator Loss: 1.6570
Epoch 1/8... Discriminator Loss: 0.7453... Generator Loss: 6.4959
Epoch 1/8... Discriminator Loss: 0.4210... Generator Loss: 4.2125
Epoch 1/8... Discriminator Loss: 1.1911... Generator Loss: 0.5131

Epoch 1/8... Discriminator Loss: 0.1382... Generator Loss: 4.3639
Epoch 1/8... Discriminator Loss: 0.1899... Generator Loss: 3.0533
Epoch 1/8... Discriminator Loss: 0.1866... Generator Loss: 2.8901
Epoch 1/8... Discriminator Loss: 0.1528... Generator Loss: 5.3423
Epoch 1/8... Discriminator Loss: 0.5672... Generator Loss: 1.2334
Epoch 1/8... Discriminator Loss: 0.7145... Generator Loss: 0.9349
Epoch 1/8... Discriminator Loss: 0.5023... Generator Loss: 1.1933
Epoch 1/8... Discriminator Loss: 2.2068... Generator Loss: 0.1833
Epoch 1/8... Discriminator Loss: 3.1436... Generator Loss: 0.0692
Epoch 1/8... Discriminator Loss: 0.7465... Generator Loss: 0.9426

Epoch 1/8... Discriminator Loss: 0.3620... Generator Loss: 2.4173
Epoch 1/8... Discriminator Loss: 1.0411... Generator Loss: 0.6702
Epoch 1/8... Discriminator Loss: 1.1707... Generator Loss: 2.4056
Epoch 1/8... Discriminator Loss: 0.5322... Generator Loss: 1.4275
Epoch 1/8... Discriminator Loss: 0.7877... Generator Loss: 0.9818
Epoch 1/8... Discriminator Loss: 1.6884... Generator Loss: 3.4219
Epoch 1/8... Discriminator Loss: 0.6833... Generator Loss: 3.9096
Epoch 1/8... Discriminator Loss: 3.0672... Generator Loss: 0.0734
Epoch 1/8... Discriminator Loss: 4.1506... Generator Loss: 0.0269
Epoch 1/8... Discriminator Loss: 0.2474... Generator Loss: 2.4365

Epoch 1/8... Discriminator Loss: 0.3123... Generator Loss: 4.2521
Epoch 1/8... Discriminator Loss: 1.9245... Generator Loss: 0.2365
Epoch 1/8... Discriminator Loss: 0.2870... Generator Loss: 2.0399
Epoch 1/8... Discriminator Loss: 1.3820... Generator Loss: 0.5568
Epoch 1/8... Discriminator Loss: 1.4916... Generator Loss: 0.3514
Epoch 1/8... Discriminator Loss: 1.5202... Generator Loss: 0.3140
Epoch 1/8... Discriminator Loss: 1.4658... Generator Loss: 0.4964
Epoch 1/8... Discriminator Loss: 0.9052... Generator Loss: 0.7974
Epoch 1/8... Discriminator Loss: 2.6454... Generator Loss: 0.0901
Epoch 1/8... Discriminator Loss: 1.1496... Generator Loss: 0.5373

Epoch 1/8... Discriminator Loss: 1.2229... Generator Loss: 2.0077
Epoch 1/8... Discriminator Loss: 0.7854... Generator Loss: 0.8929
Epoch 1/8... Discriminator Loss: 0.9220... Generator Loss: 0.6328
Epoch 1/8... Discriminator Loss: 0.6565... Generator Loss: 1.0531
Epoch 1/8... Discriminator Loss: 0.2237... Generator Loss: 2.7773
Epoch 1/8... Discriminator Loss: 4.4346... Generator Loss: 4.8701
Epoch 1/8... Discriminator Loss: 0.2814... Generator Loss: 2.3953
Epoch 1/8... Discriminator Loss: 0.3438... Generator Loss: 5.4863
Epoch 1/8... Discriminator Loss: 0.2502... Generator Loss: 3.7980
Epoch 1/8... Discriminator Loss: 1.7199... Generator Loss: 2.7501

Epoch 1/8... Discriminator Loss: 0.4549... Generator Loss: 3.1461
Epoch 1/8... Discriminator Loss: 1.0708... Generator Loss: 0.5267
Epoch 1/8... Discriminator Loss: 0.5992... Generator Loss: 0.9737
Epoch 1/8... Discriminator Loss: 0.6233... Generator Loss: 2.4113
Epoch 1/8... Discriminator Loss: 0.8281... Generator Loss: 2.5797
Epoch 1/8... Discriminator Loss: 0.4606... Generator Loss: 6.6175
Epoch 1/8... Discriminator Loss: 1.4279... Generator Loss: 0.3520
Epoch 1/8... Discriminator Loss: 1.2971... Generator Loss: 0.5073
Epoch 1/8... Discriminator Loss: 0.5789... Generator Loss: 1.4037
Epoch 1/8... Discriminator Loss: 0.4341... Generator Loss: 1.5570

Epoch 1/8... Discriminator Loss: 0.3856... Generator Loss: 5.3544
Epoch 1/8... Discriminator Loss: 3.6989... Generator Loss: 3.3202
Epoch 1/8... Discriminator Loss: 0.4679... Generator Loss: 1.3811
Epoch 1/8... Discriminator Loss: 1.3028... Generator Loss: 0.3876
Epoch 1/8... Discriminator Loss: 1.0926... Generator Loss: 0.5679
Epoch 1/8... Discriminator Loss: 0.7440... Generator Loss: 1.8650
Epoch 1/8... Discriminator Loss: 0.3616... Generator Loss: 2.9253
Epoch 1/8... Discriminator Loss: 0.5196... Generator Loss: 1.3852
Epoch 1/8... Discriminator Loss: 1.0198... Generator Loss: 0.5737
Epoch 1/8... Discriminator Loss: 0.8608... Generator Loss: 0.7758

Epoch 1/8... Discriminator Loss: 0.3279... Generator Loss: 5.0152
Epoch 1/8... Discriminator Loss: 0.3534... Generator Loss: 1.8591
Epoch 1/8... Discriminator Loss: 0.8468... Generator Loss: 1.0670
Epoch 1/8... Discriminator Loss: 1.1815... Generator Loss: 0.4859
Epoch 1/8... Discriminator Loss: 0.3361... Generator Loss: 1.5991
Epoch 1/8... Discriminator Loss: 0.8955... Generator Loss: 1.7185
Epoch 1/8... Discriminator Loss: 1.1942... Generator Loss: 1.0050
Epoch 1/8... Discriminator Loss: 0.5614... Generator Loss: 1.6181
Epoch 1/8... Discriminator Loss: 0.9317... Generator Loss: 0.7038
Epoch 1/8... Discriminator Loss: 0.3146... Generator Loss: 2.2593

Epoch 1/8... Discriminator Loss: 1.7013... Generator Loss: 0.3028
Epoch 1/8... Discriminator Loss: 0.8825... Generator Loss: 1.1154
Epoch 1/8... Discriminator Loss: 1.0132... Generator Loss: 0.8399
Epoch 1/8... Discriminator Loss: 1.0572... Generator Loss: 0.6166
Epoch 1/8... Discriminator Loss: 2.5846... Generator Loss: 0.1024
Epoch 1/8... Discriminator Loss: 0.5234... Generator Loss: 2.2622
Epoch 1/8... Discriminator Loss: 0.4678... Generator Loss: 3.4961
Epoch 1/8... Discriminator Loss: 0.2039... Generator Loss: 3.5506
Epoch 1/8... Discriminator Loss: 0.3395... Generator Loss: 2.6504
Epoch 1/8... Discriminator Loss: 2.8299... Generator Loss: 0.0840

Submitting This Project¶

When submitting this project, make sure to run all the cells before saving the notebook. Save the notebook file as "dlnd_face_generation.ipynb" and save it as a HTML file under "File" -> "Download as". Include the "helper.py" and "problem_unittests.py" files in your submission.